Hi Recently, after few weeks of tests of Nautilus on our clusters we decided to upgrade our oldest one (installed in 2012 as bobtail release). After gateway upgrade we found that only for some buckets (40% from ~2000) the same request is handled differently. With mimic RGW - OK (200), with nautilus RGW - Bad request (400). All the requests (GET, HEAD, PUT) for particular bucket are refused, for other buckets are handled correctly. So currently we are running nautilus cluster with mimic gateways to bypass the problem.
All the failing buckets uses wrong index in explicit_placement settings. We are using two storage classes and those buckets contains the same *data_pool* and *index_pool*: *.rgw.buckets* where we are using *.rgw.bucekts.index. * Also *Data extra pool* field is always empty for those buckets. I've checked and *.dir.<bucket_id>* objects for this buckets are stored on data pool, not index pool - so according to *explicit_placement* settings How can I move them with all metadata keys and update buckets settings? Where information about *explicit_placement* is stored? Additionally, is there a way to find out when bucket has been created? *radosgw-admin bucket stats* command do not provide it, but maybe it's available based on a timestamp of some objects or their keys? Jacek -- Jacek Suchenia [email protected]
_______________________________________________ ceph-users mailing list [email protected] http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
